Ahmed Hefnawi (Volta) currently i am going to participate in a hackathon and i have been working on the similar idea also .

Currently my team is trying to analyze by bioinformatics tool the optical symmetries , as u know that already the virus is proved to be Raman positive and detected in a very well manner ; so there is a chance that it may be positive for IR or may not based on molecular rather centroid symmetry.

Data is very much needed by experiments on known positive samples with IR spectrometer and then we should try to minimize the cost and size of the spectrometer and the Deep neural network for interpretation 

What are your opinions on the Public Lab Smartphone-based spectrometer? If you were to make some modifications to your camera, you might be able to get Vis-NIR spectra on your smartphone...  https://spectralworkbench.org/

That would be very interesting project to experiment with and check out.

But with the optical window required for a reliable application in virology, I'm not sure that would be a reliable option at least not during the phase of collecting correct spectra using a calibrated sensor along with NIR LEDs or Laser Diodes. What do you think?

Potentially - there are some papers that mention the 650-1050 range being adequate for such thing - this would not be done on a phone but something more along the lines of OpenMV, etc.

Yes exactly, that is the optical window that is quoted by most of the published papers also in my list above.

That's interesting, I wonder how would an OpenMV handle NIR and would be interested in testing it after establishing the spectra, and I think it might not be a match for specialised sensor with a decent spectral response range of 640 - 1050 nm, not only that but a spectral resolution of 17-20 nm, but still could be used in different other applications outside of virology.

Also, I wonder how would the Raspberry Pi NoIR camera would do as well, endless possibilities once we have established spectra and then I would be very interested in making comparisons between OpenMV, RPi NoIR, etc against the readings of the spectrometer sensors.

Would you be interested to work on this project with me? As per my latest update I don't have enough time to proceed so I'm openly sharing every bit of information I have on this project so that others can pick up and move it forward :)

Thanks,

Ahmed

Hey Ahmed, 

I would be really excited to work with you on this, and sorry for the late reply. I am currently talking with someone who is working on implementing NIRS for real-world devices, and he also seems interested. Would you mind emailing me (vdaita@gmail.com) and putting me in touch with anyone else you know working on NIRS-based COVID-19 detection? I might be able to help them with their research.

Also, quick sidenote, for a hackathon, I gathered a team that was able to build a model that could analyze COVID-19 Raman spectra using PCA, and we are planning to add noise reduction.

Thank you,

Vijay

Thank you Vijay for your reply.

That sounds very interesting, I will drop you an email and would be looking forward to collaborating with you or your team.

The analysis of Raman spectra using PCA sounds like a good start regarding your hackathon, let me know also if I can lend a hand at any time over weekends, I'm currently a bit busy with other stuff, but can always spare an hour or two every weekend.

Thanks,
Ahmed

How are you differentiating in the spectral domain ? Are you using a diffraction grating and precision optics?  Or maybe you have a narrowband tunable wavelength infrared source?  This is the single most important aspect of the hardware design.   I would say easily trumps over the choosing an infrared sensor.  

STOP PRESS - EDIT   I've just seen this 18 slot channelised spectrometer on here    https://hackaday.io/project/143014-compact-25-spectrometer

For example the last spectrometer I did was 30 years ago fixed band only looking at one pair of absorption wavelength. It had used precision thinfilm narrow bandpass optical filters built at 185 nm and (??can't remember nm) spot frequencies. Optical design at UV was a challenge because ordinary glass couldn't be used , not much a problem at nearinfrared.     

I expect that once there's a prototype detecting things OK, interest and serious support will then really kick off.      

Thank you so much for the suggested spectrometer project, there are a number of open source spectrophotometer designs as well available and one of them in my listed literature. I would say they are a very helpful source. Specially the AS7265x chip, I will look into it indeed!

Speaking of the spectral domain, I'm not planning on using diffraction grating, I want this prototype to be easy to build and in the future easy to manufacture. I would go with a very low spectral bandwidth of less than 30nm probably play with some laser IR emitters for various wave lengths in the range of 650nm to 1000nm which is the most useful range for biological samples called the "optical window" as suggested by the literature as well. Anything less than 650nm or greater than 1000nm will be greatly distorted by the high absorption of water and if we are using red blood cells by hemoglobin. And yes, I will be working with near-infrared at least during the current phase, things could change in the future.

That's great to know :) Yeah, I'm looking forward to having a prototype that can produce very reliable spectra and would really appreciate your comments and suggestions along the way since you have already worked with spectrometer.

Stay tuned and have a great day!

Ahmed

The literature is not about SARS-CoV-2 but one of the researches tackled HIV-1.

There is no exact fingerprint for the spectra of the SARS-CoV-2 available yet, but this will be extracted from the spectra and data analysis done on samples after constructing the first prototype. Also, the protein structure of the virus is now available and I can use this to infer the fingerprint, but again it will depend on the sensitivity we can reach with the sensor and the spectra produced.

The plan is to test samples from PCR or RT-PCR confirmed patients (both positive and negative) and start correlating the output from the spectrophotometer.
Thank you for asking and I would be happy to hear back from you.

Can you give a quick summary of the literature? What's the signature that you'll be looking for that's specific to SARS-CoV-2 in an infected patient?

Wrt literature summary, I'm writing my thoughts from all the literature that I have gone through so far, and hope it helps. Also please feel free to suggest or ask or comment on any future project log or anything that doesn't seem right. I'm very open to feedback and ideas.